Platform
The platform is structured around four capabilities. Each layer composes the next: observation feeds intelligence; intelligence constrains autonomy; autonomy validates against simulation; and simulation feeds back into observation as ground-truth for the next iteration of every model.
01
Multimodal ingestion — every signal an energy system emits.
Telemetry from PMUs, AMI, SCADA, IEC 61850 IEDs, IEEE 2030.5 inverters, BACnet BMS, Matter-bound residential devices, smart panels, and EV chargers. Vision and audio from drones, pole cameras, and acoustic sensors. Weather and market feeds. Standards corpora. All time-aligned, all addressable.
How observation works02
Foundation models, fused into one understanding.
TimesFM, Chronos, and Moirai for forecasting load, generation, EV charging, and heating demand. Qwen3-VL and Pixtral for inspection. PANNs and YAMNet for acoustic anomaly detection. GraphCast for weather. Gemini for reasoning and tool orchestration. Embeddings and long-term memory tied to each site under management.
How intelligence works03
Agentic action, with auditable provenance.
An agent runtime that calls forecast, simulate, dispatch, and search tools. VPP bidding into FERC 2222 markets. Demand response over OpenADR 3.0. FLISR augmentation. Microgrid islanding. EV smart-charging schedules. Every decision logged with inputs, model lineage, and approval policy.
How autonomy works04
Test the future, before you wire it.
Physics-aware power-flow and DER orchestration on OpenDSS / GridLAB-D foundations, plus building-energy MPC and 5GDHC thermal models. The same simulator powers the autonomous dispatch loop in production — the agent is constrained to actions that simulate cleanly first.
How simulation works